Re: c/7464: weak symbols alias definitions are broken in 3.1.1 (weak alias regression from 2.95)

Re: c/7464: weak symbols alias definitions are broken in 3.1.1 (weak alias regression from 2.95)

[Andreas Jaeger]

The following reply was made to PR c/7464; it has been noted by GNATS.

From: Andreas Jaeger <aj@suse.de>
To: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
Cc: gcc@gcc.gnu.org, <gcc-gnats@gcc.gnu.org>
Subject: Re: c/7464: weak symbols alias definitions are broken in 3.1.1
 (weak alias regression from 2.95)
Date: Fri, 25 Oct 2002 16:25:26 +0200

 Wolfgang Bangerth <bangerth@ticam.utexas.edu> writes:
 
 > This is one more regression from 2.95 with respect to weak and alias 
 > attributes:
 > --------------------------------------------------
 > extern int i __attribute__ ((alias ("__i")));
 > extern int i __attribute__ ((weak, alias ("__i")));
 > --------------------------------------------------
 >
 > In contrast to c++/6074, which is similar but treats functions instead of 
 > variables, this is broken in both C and C++, and is also not fixed on 
 > mainline.
 >
 > The messages I get are:
 > from C:
 >     x.c:2: error: weak declaration of `i' must precede definition
 > from C++:
 >     x.cc:2: error: redefinition of `int i'
 >     x.cc:1: error: `int i' previously declared here
 >
 > As said, gcc2.95 accepts this.
 
 But that doesn't make it right here.  The warning is correct.
 
 Andreas
 -- 
  Andreas Jaeger
   SuSE Labs aj@suse.de
    private aj@arthur.inka.de
     http://www.suse.de/~aj

Re: c/7464: weak symbols alias definitions are broken in 3.1.1 (weak alias regression from 2.95)

[Richard Henderson]

The following reply was made to PR c/7464; it has been noted by GNATS.

From: Richard Henderson <rth@redhat.com>
To: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
Cc: gcc@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Subject: Re: c/7464: weak symbols alias definitions are broken in 3.1.1 (weak alias regression from 2.95)
Date: Fri, 25 Oct 2002 11:24:00 -0700

 On Fri, Oct 25, 2002 at 09:08:25AM -0500, Wolfgang Bangerth wrote:
 > This is one more regression from 2.95 with respect to weak and alias 
 > attributes:
 > --------------------------------------------------
 > extern int i __attribute__ ((alias ("__i")));
 > extern int i __attribute__ ((weak, alias ("__i")));
 > --------------------------------------------------
 
 This test case is broken.  Despite the use of "extern",
 the use of "alias" means that this is a definition.
 So having two of them violates the ODR, and is indeed an error.
 
 
 r~

Re: c/7464: weak symbols alias definitions are broken in 3.1.1 (weak alias regression from 2.95)

[Wolfgang Bangerth]

The following reply was made to PR c/7464; it has been noted by GNATS.

From: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
To: Andreas Jaeger <aj@suse.de>
Cc: gcc@gcc.gnu.org, <gcc-gnats@gcc.gnu.org>
Subject: Re: c/7464: weak symbols alias definitions are broken in 3.1.1 (weak
 alias regression from 2.95)
Date: Fri, 25 Oct 2002 09:39:13 -0500 (CDT)

 > > --------------------------------------------------
 > > extern int i __attribute__ ((alias ("__i")));
 > > extern int i __attribute__ ((weak, alias ("__i")));
 > > --------------------------------------------------
 > >
 > > The messages I get are:
 > > from C:
 > >     x.c:2: error: weak declaration of `i' must precede definition
 > > from C++:
 > >     x.cc:2: error: redefinition of `int i'
 > >     x.cc:1: error: `int i' previously declared here
 > 
 > But that doesn't make it right here.  The warning is correct.
 
 OK, then we should close the report, right?
 
 Regards
   Wolfgang
 
 PS: I still find the messages confusing -- both C and C++ talk about 
 "definitions", but there are none in this case...
 
 -------------------------------------------------------------------------
 Wolfgang Bangerth              email:           bangerth@ticam.utexas.edu
                                www: http://www.ticam.utexas.edu/~bangerth
 
 

Re: c/7464: weak symbols alias definitions are broken in 3.1.1 (weak alias regression from 2.95)

[Wolfgang Bangerth]

The following reply was made to PR c/7464; it has been noted by GNATS.

From: Wolfgang Bangerth <bangerth@ticam.utexas.edu>
To: gcc@gcc.gnu.org, <gcc-gnats@gcc.gnu.org>
Cc:  
Subject: Re: c/7464: weak symbols alias definitions are broken in 3.1.1 (weak
 alias regression from 2.95)
Date: Fri, 25 Oct 2002 09:08:25 -0500 (CDT)

 This is one more regression from 2.95 with respect to weak and alias 
 attributes:
 --------------------------------------------------
 extern int i __attribute__ ((alias ("__i")));
 extern int i __attribute__ ((weak, alias ("__i")));
 --------------------------------------------------
 
 In contrast to c++/6074, which is similar but treats functions instead of 
 variables, this is broken in both C and C++, and is also not fixed on 
 mainline.
 
 The messages I get are:
 from C:
     x.c:2: error: weak declaration of `i' must precede definition
 from C++:
     x.cc:2: error: redefinition of `int i'
     x.cc:1: error: `int i' previously declared here
 
 As said, gcc2.95 accepts this.
 
 Regards
   Wolfgang
 
 -------------------------------------------------------------------------
 Wolfgang Bangerth              email:           bangerth@ticam.utexas.edu
                                www: http://www.ticam.utexas.edu/~bangerth